DESECHO DESCRIPCIÓN Aceites, lubricantes y

Extension-oriented participants stressed the benefits of extending general scientific knowledge of climate variability in terms of mechanistic and statistical associations (Chapter 2). Wetter and drier periods in the instrumental record have occurred at various timescales; in histories of rural Australia, drought and deluge appear as punctuation in the lives of industries and communities (Chapter 2). Thus, even being able to describe the climate of the rangelands statistically recreates anomalies such that extreme events no longer appear as things which have happened, but rather as things that do happen. The over-layering of rainfall records with the mechanistic concepts such as ENSO, teleconnection, the Madden Julian Oscillation (MJO), and others creates a conceptual climate that can renovate the sense of uncertainty that has historically pervaded understandings of what the future holds, climatically speaking. There has been substantial work done by climate risk technologists in extending these conceptual aspects of climate. Together they redefine what might be deemed appropriate for the future of Australian agricultural and ecological management. Nonetheless, as I will go on to argue, the complexity of some climate-related boundary objects means that extending the conceptual climate can create translation risks.

Extending mechanism

Participants acknowledged that agricultural decision-makers are diverse and do not just want one thing. Some graziers want information, others want

knowledge. Some want all the theory, others just want the advice that has been built from a synthesis of theory, empirical research, modelling and practice. Graziers’ country and their businesses are also diverse and diversely understood (Ison and Russell 2000). Nevertheless, among the participants, those who had actively engaged with agricultural decision-makers frequently suggested that application of seasonal climate information required some understanding of the mechanisms by which weather and climate were generated and thus predicted. These participants spoke of the need to clarify an apparent contradiction between the inability of scientists to predict rainfall more than seven days out while claiming to be able to forecast seasonal rainfall for the upcoming three months. Thus, explaining the mechanisms of ENSO and its Australian teleconnections was described as a necessary grounding for establishing public credibility of statistical forecasts. Such explanation often proceeds from descriptions of the weather systems that bring rain (e.g. cyclones, fronts, monsoonal troughs) and how these can be affected by different ENSO conditions. In itself, garnering a basic mechanistic understanding of climate via weather was described as an absolutely necessary step in making the very idea of climate prediction tenable to decision-making. As one extensionist put it:

You've got to make the connection between a warm patch of ocean in the Pacific and rain at somebody's backyard, and without the weather we couldn't do that, so we had to put in the weather stuff, yeah, and that's still the case (R31).

According to this line of thinking, it is by explaining mechanism and local impact that credibility of climate forecasting is built in different places. Explicating weather processes and climate as the sum of those processes pays dividends in seeing ‚people’s eyes light up‛ (R31). Such descriptive work was seen as a preparing people to more readily engage with seasonal climate forecasting; it was equally regarded as equipping decision-makers with tools to interpret weather maps, to reconfigure their knowledge of their agro-ecosystems in climatic terms and to enter into the language and semiotics of climate and weather.

Extensionists description of the need for good extension indicates that models and forecasts do not gather a great following by themselves; application is predicated by an explanation of why climate is predictable. The mechanistic aspects of climate are particularly important in mediating probabilistic forecasts which, because they are poorly described in terms of hits and misses, make public validation problematic. That is, participants were keen to discourage people from judging forecasts as ‘right’ or ‘wrong’ because this language jars with the probabilistic outputs of forecasts. Without recourse to personal and local validation, as might be found in technologies which can be trialled or adapted to local conditions, the explanation of mechanism was described as a crucial step in the process of adoption climate science among agricultural publics. Mechanism was backed up by history-as-statistics as a measure of reliability.

Building futures out of climatic history

A mechanistic association may be at least partially explanatory; historical association, can be convincing. A major effort for climate risk technologists has been to extend analyses of climatic history as a means of demonstrating the strength of associations between rainfall and ENSO indices. An emphasis on instrumental histories of local and regional rainfall constitutes ‘climatology’ as partitioned into ENSO states. Where probability distributions of the past are used to represent climatology, the degree to which the indices of ENSO explain the variability across history creates a visual sense of teleconnections. Some extensionists argued that this statistical and historical view of climate, coupled with a basic mechanistic understanding of what makes regional rainfall, can change people’s conception of their rain-fed agriculture and interest them in the possibilities of incorporating climate forecasts and other scientific climate analyses in their decision-making. The premise of this argument is that what happened in the past is a useful guide to what will happen in the future. Thus, statistical representation tends to recast the future as a probabilistic duplication of the statistical past. Although there is some awareness of the problems associated with such a frequentist view among climate risk technologists (e.g.

representations of the past into probabilistic representations of the future for publics; that is to translate a statistical history (probability distribution) into a forecast (probability). For example, the DSS, Australian Rainman (Clewett et al. 2003), generates both tables and graphs of historical probability distribution, and these express the historical distribution simply as ‘probabilities’ (see Figures 7.4, 7.5 and 7.6 for examples). In many cases, such a practice would appear to be a matter of convenience, a form of shorthand. Nonetheless, it is a particular framing of climate variability that emphasises the known over the unknown. The odds or probabilities of a given amount of rainfall in such situations were portrayed by some climate risk technologists as a means of making decisions easier through formally rationalising the process by which they are made. As one researcher explained:

there's some exercises where you get people to, you know, write down options and look at even simple probabilities [which] actually I think can be liberating for them sort of thing in terms of that. < I think it's often a useful way of thinking about the uncertainty < and being probably explicit about what it is. < I think some notion of rainfall and seasonal forecasts and so on < can be an empowering thing to think about. ‚Well we don't know what's happening in the future, but these are the odds of going this way or that way.‛ Also so that people don't beat up on themselves too much if things do go wrong because they still know that, well, that was probably the right decision at that time (R30).

This account is an informative one; it is explicitly concerned with the well-being of decision-makers, not just their decisions. Calculative rationality provides for the ability to externalise or neutralise blame for results of decisions by emphasising the apparent objectivity of using ‘the best available information’, weighed up in the most sensible manner possible. The right decision on the basis of the available information recreates the local decision and decision-maker within a risk assessment framework. This script of ‘the best possible information’ was one that was commonly invoked by climate risk technologists as they explained the usefulness of probabilities.

The indices of ENSO themselves have also become important objects by which climate variability can be conceptualised by agricultural publics. Early climate

extension to Queensland graziers provided instruction on do-it-yourself SOI phase system forecasting. Graziers were shown how to compare their own property rainfall records to the SOI and its phases in the historical record. However, this simple and local way of forecasting is riddled with artificial skill (e.g. Barret 1998). Its strength, though, was that it gave individual managers or families a view of the bearing of the SOI on their rainfall. The relationship between ENSO indices and the local dry or wet periods could be witnessed through correlations. When the indices are used to construct ENSO states, such correlations take on the more tangible form of analogue years, which allow agricultural publics to look at history as something that can be categorised in terms of ENSO. Thus ENSO indices and states were provided to publics as a means of understanding the local impacts of ENSO by drawing on people’s own records, and leading to a re-analysis of historical, lived variability in terms of ENSO. As one applications researcher put it:

The beauty of packages like Rainman is that you can say, "For your location over the last 100 years, this has been the effect of ENSO", and they can look at the data. They can just go through it slowly, year by year, and they can see how good the relationship is, and how poor it is without [ENSO]. As you move into the future < you move away from that ability – that's a real downside of the modelling. So believability and credibility < will be at risk (R17).

In Queensland, the SOI is commonly held up against historical records allowing a new predictive meaning to be found in property rainfall records. The commitment to the SOI as an index of ENSO is highlighted by its weekly appearance on the weather segment of the ABC television news in Queensland. The primacy of the SOI is also evident in the way Rainman was sometimes discussed. As well as providing a means to conduct one’s own analysis of climate data from stations across Australia, Rainman allows users to upload their own rainfall data and develop place-specific SOI-related probability distributions, which can be easily mistaken for forecasts. Thus, ENSO can become locally credible as an element in decision-makers assessments by being made visible through the SOI in ways climate scientists would not consider credible. In the making of a place-specific public science, the ontology of climate science is

eroded, as is the technical credibility associated with skill. In its place, the salience of locality, and the legitimacy of self-assessment comes to play a more important role in the production of a different epistemic form of credibility. The situational, in this scenario, rules over the universal, and application becomes an instrumental driver which can outweigh the cautions of the climate scientists about the pre-eminent importance of skill and artificial skill. While skill is still attended to in Rainman, the spatial coherence that is a pre-requisite of forecast skill is cast aside in favour of local relevance and ‘ownership’.

From interpretation of the quasi-global indices to make risk judgements about local impacts, a picture emerges of the boundary objects of climate science being extended across all manner of boundaries in ways that were not first intended by climate scientists. Knowledge-making practices are morphed through the production of boundary objects which are perhaps tailored more to the pre- requisites of the local climate and thus the places in which risk assessments are made. Scientific knowledge is repackaged as information and then re-integrated in locales by individuals, families or groups with different epistemic cultures and means of producing knowledge, to influence decision-making in unknowable ways.

Trends, steps and the framing of credibility and salience

Climate is neither stationary nor immune to regional and wider scale trends. Changes in the global climate have been linked to substantial regional changes in the rainfall or temperature climatology of particular places (e.g. IOCIP 2002). In some instances, the strength of the relationship between climatic variables and anthropogenic climate change appears to easily cement trends as fixtures of climate for current and future agricultural management. For instance, the global tendency for night time temperatures to be increasing faster than day time temperatures has been uncontroversially fixed to theoretical understandings of the action of enhanced greenhouse conditions. The narrowing of frost windows in the northern grains belt of Queensland is easily stabilised as an ongoing trend (see Meinke and Stone 2005). Stochastic inter-annual variation in the frost

window is expected, but in this case the trend becomes a dominant feature of historical analysis.

Rainfall trends and steps have been harder to pin to anthropogenic climate change. These changes may be temporary aberrations associated with climate variability at decadal, inter-decadal or even longer timescales. Predictability at such timescales is a subject of ongoing research within the climate science community who remain largely unsure of the mechanisms associated with particular shifts (Chapter 5, Section 2). For example, in the south-west of Western Australia, a step-like decline in winter rainfall commenced in the early to mid 1970s, but was not fully recognised for the best part of three decades (IOCIP 2002). While this shift accompanied other major changes in the climate system globally, the stability of the new rainfall regime cannot be inferred (Nicholls 2003). The rainfall for this region (or any other) may change again and perhaps decrease even further. Shifting the temporal bounds of what is included in ‘climatology’ (i.e. to include only the last 30 years) is a stop-gap measure, the adequacy of which is largely indeterminate.

This scenario presents issues requiring careful balance for climate risk technologists in performing climate variability to lay audiences and policy- makers. On one hand, the usefulness of the instrumental record as a basic (though far from perfect) tool is eroded. The notion that climate is unstable and, moreover, that it changes in non-linear ways, starts to undermine probabilistic forecasts; it is suggestive that probabilities do not ‘speak for themselves’. The unsettling understanding that regional climate does not always change in linear ways overlays climate prediction with unwieldy indeterminacy and ignorance. Where stability is belied by the emergent features of climatic distributions and unprecedented and unusual events, the ‘best available knowledge’ needs cautious language to translate its complexity into simple terms. As one climate scientist put it: ‚It would be easier to make seasonal forecasts if the ‘background’ climate was not changing so rapidly. I suspect that this is changing the relationships on which we base our forecasts‛ (R2).

Human action in this story-line appears to be dismantling the stability of climate associations. Climate forcings – which are now both ‘social’ and ‘natural’ – reconstitutes seasonal climate risk as a late modern risk (Beck 1992, and see Chapter 3). Alternatively, taking Ravetz’s (1986; 2005) conception of uncertainty in science, as climate change science has become ‘post-normal’ (Bray and von Storch 1999) it has started to disturb the supposedly ‘normal’ science of ENSO (e.g. Thornes and McGregor 2003) by placing the stability of historical associations in doubt. These concerns frame a substantial challenge for climate risk technologists: how should they translate multiple forms of uncertainty associated with climate variability without entirely undermining the salience, credibility and legitimacy of their forecasts?

It would appear that making climate prediction requires substantial rhetorical resources to frame it such that the associated ‘uncertainties’ are made visible, yet not made to swamp the potential value to decision-makers. The separation of signal from noise is a perpetual challenge. When an empirical signal cannot be mechanistically settled among climate scientists, the gatekeepers of climate at ABoM were reluctant about letting these elements of climate science out of the ‚research bag‛ (R4) and into the public domain. In contrast, some applications researchers and extensionists appeared more bullish in their suggestions of the potential for agricultural decision-makers to apply climate science, even if it is not fully understood (cf. Shackley and Wynne 1996). One extensionist referred to a ‚policy of perfectionism‛ (R14) among climate scientists at ABoM. In contrast, their own approach might be described as a ‘policy of pragmatism’. This participant exemplified the tension between the two ‘policies’ in relation to the Madden Julian Oscillation (MJO), or as Queensland climate risk technologists have packaged it, the 40-day Wave:

The Bureau didn't want me to talk about it and yet people [farmers/graziers] here are using it because they say, "Look there's one supposed to be coming up in a couple of weeks' time, I will wait for that" or "It's come and it's gone and it didn't rain. I know there's very little chance of getting any more good rain in the next three weeks". So that's a tactical decision and I think it's a very useful one. < Everybody knows that it is not [a] regular 40 days. < We used to popularly call it the 40-day Wave here, and then the Bureau said

"No, you've gotta call it a 30/50 [day wave]". Then it was a 30/60 and then < ‚You mustn't use that term, you're going to make it all confusing. [Call it] an intra-seasonal oscillation or an MJO‛. But the 40-day Wave says: "Every six weeks approximately this [increased chance of rain] is happening", and again it's this business – the policy of perfectionism. Do you put out something to farmers that is catchy? ‚A 40-day Wave – oh, we all remember that

– or what's the name – JAMA, umm, inter-what?‛ < My job as you know < is

turning science into something that for people is easy to understand and that they want to read about and you run into the problem of the scientist who always wants to qualify every statement you make and that is a problem < if you start putting all the qualifications in < you can lose the message (R14).

From this quote, the policies of perfectionism and pragmatism can be seen as divergent takes on the construction of scientific boundary objects, the former emphasising scientific credibility, the latter its salience to decision-makers. According to the policy of perfectionism, scientific credibility of public knowledge is the only issue in deciding what should be made public. For climate scientists, difficulties in estimating skill, of understanding mechanisms, and the broad, highly social work of consensus building around forecasts and noisy phenomena are the primary elements in the making climate information into climate applications.

What climate scientists consider as insufficiently understood phenomena are regarded very differently by some applications researchers. Their pragmatic approach to building boundary objects attempts to balance uncertainties that